The Panel Study
IN THIS BOOK, we use two main sources of data to study choice behavior and charter schools. One dataset, which we describe in chapter 5, is a record of the characteristics and electronic search behavior of parents using an Internet site, DCSchoolSearch.com, that we created to help parents shop for schools. In this chapter, we describe the other major source of data, a four-wave panel survey in which we interviewed at four separate times a sample of Washington, D.C. parents with children either in the charter schools or in the traditional public schools.1 In the last two waves, we interviewed students as well.2
We conducted the first wave of parent interviews in the Fall of 2001, at which time we talked with just over five hundred parents in charter schools and about the same number of parents whose children were in traditional public schools.3 At the end of that school year (Spring 2002), we reinterviewed as many of these parents as possible. As we will see below, attrition was high even over the course of a single school year and in Wave 2 we completed interviews with just about six hundred of the parents in our original sample. Wave 3 was conducted in the Fall of 2003, when we successfully reinterviewed 395 of the original parents. In each wave of the parent survey, we asked the parent to talk about a specific child and the specific school in which that child was enrolled.4
In Wave 3, we asked the parent for permission to speak directly with the student if that child was enrolled in grade 7 or above. We completed interviews with 195 students. Of these, 100 were in charter schools, while the other 95 were in traditional public schools. At the end of the 2003/4 school year we launched Wave 4, during which we reinterviewed 297 parents and 149 students.
There are several important characteristics of this component of our research. First, of course, our study employs a panel—a research design that is all too rare in the social sciences. As in any panel study, the structure of the data allows us to measure both stability and change in attitudes and behavior over time. This is always important when investigating the effects of institutions on individuals, but is particularly important when investigating a relatively new institutional reform, such as the charter schools we